Cost per authorisation decision
Continuous assurance is not free. But a point-in-time decision, made once on evidence that is already stale, is dearer than it looks once you price the blind spot it leaves behind. Set the classification, choose a model, and let an AI-assist take some of the manual load. The human decision stays human.
Drive the model
Higher classification means more controls and tighter handling: more hours per control.
One big re-accreditation every 3 years, or a steady continuous cadence.
Share of the manual evidence and drafting work (crosswalks, POA&M, summaries) an assistant takes off the human. The decision stays human.
Fully loaded rate derived from the grade's salary midpoint. The decision stays human.
What this shows. A point-in-time re-accreditation is one big batch that buys a single decision, so its cost-per-decision is tall and amber. A continuous model spreads a lean, mostly automated effort across many decisions, so each one is cheap and teal. AI-assist bends the manual portion down in both; because continuous is already lean, the saving is a bigger share of a smaller bill.
What this shows. Point-in-time evidence ages in a long ramp: for most of the cycle the certificate asserts a posture that is over a year old, and the red area above the stale line is the blind spot you are carrying. Continuous re-verification keeps evidence young and the blind spot bounded to one cadence.
Illustrative. Synthetic figures, not a costing tool. A deterministic teaching model with hand-authored constants: the picture is identical on every load. The authorisation decision stays human.
Rates derived from the ASD Non-SES Salary Scales (from 1 September 2025), loaded at ~1.7× base (incl. 15.4% employer super) over ~1,800 hours. Illustrative.